Chapter 5 Least-cost paths

In this chapter we use the ‘least cost paths’ (a.k.a. LCP) methodology defined in Chapter 3 to define the path of least resistance from all coastal protected areas (larger than 1km2) in the focal region to their corresponding high elevation protected area. For each protected area (Section 1) and mangrove (Section 2), we calculate the LCP for the 10 nearest (by euclidean distance) high elevation protected areas, then select the path with the least resistance. We then explore several important properties of these paths between different countries (e.g. path length, mean resistance, total resistance, quality of forest on the path), in order to understand the spatial state of low-to-high elevation connectivity in central America. Finally, we devise a prioritization scheme to select candidate locations for habitat restoration initiatives.

It is important to note that the LCP method does not fully characterize connectivity in a probabilistic fashion, but it is a tool which cn give us simple information to design prioritization schemes. For a more complete consideration of connectivity, see the next chapter.

5.1 ‘End’ Nodes

We first start by defining a suite of end points (or nodes) - high elevation protected areas. To do this we import all of the high elevation protected area polygons defined in the data assembly section. As the LCP methods require points as targets, not polygons, we also randomly sample within the polygons to create a complete suite of “end nodes”. This means that there are multiple end points within larger protected areas.

In the map below, all of the end points are shown as black points, and their corresponding protected area polygons in blue:

In total, we have a suite (n = 1659) of end points for the analyses. For each protected area, we find the 10 closest of each of these end points (euclidian distance) and calculate the least cost paths to them.

5.2 Protected area climate corridors

We used the package ‘gDistance’ and the function ‘shortestPath’ to define the least-cost path from coastal protected areas larger than 1km2 in size to the ten nearest high elevation protected areas. Note, we use starting points adjacent to the ocean to fully encapsulate the ridge-to-reef gradients.

The map below shows the shortest (by distance) path (back lines) for each of the coastal protected areas. The underlying raster layer shows the cost surface, with red areas being highly resistant to movement, and blue areas offering low resistance.

5.2.1 Least-cost-path characteristics

In the following section we explore a suite of characteristics which would make an area a “good” candidate for restoration to preserve low to high elevation connectivity. The categories are as follows:

  • large starting area: a sufficiently large starting area would ensure strong “supply” of animals, as larger areas have higher wildlife carrying capacities.

  • large end area: a sufficiently large end location would ensure animals can actually persist in the location they arrive in again, as the carrying capacity for wildlife should be higher.

  • Short least cost path: short paths will be cheaper to restore than long paths, both in terms of the amount of restoration required, but also in terms of the logistics of tree planting.

  • Path which is already partially protected: in theory, if a path is fully protected from the start to the end, it will not need any investment in restoration. This might occur if the low elevation areas and high elevation target are within the same protected area. If this is the case, the elevation corridor should be safe from future development.

  • Path is already of good quality: it may be the case that the path travels through unprotected areas which are high quality forests, these paths would be a high priority for future conservation and resoration initatives.

-Path which doesnt cross multiple administrative borders: the more administrative boudaries (e.g. countries) a path traverses, the more logistically complex its protection and restoration will be.

In the following section, we compare the LCP characteristics between different countries to get a general understanding of the state of ridge-to-reef connectivity.

5.2.1.1 Number of protected areas available

Understanding how many different potential source populations there re per country can give an idea of where might be appropriate to start conservation initiatives. The barplot below shows the number of coastal protected areas (>1km2) by focal country. As you can see, Costa Rica has the largest number of coastal protected areas, followed by El Salvador then Panama. Belize has the least.

5.2.1.2 Starting protected area size

Not all coastal protected areas are equal however, and some have the potential be be quite small. The plot below shows the distribution of start point protected areas by country.

On average, Nicaragua has the largest coastal protected areas. Excluding the marine protected areas, the most substantial are Darien (Panama) and the Río Plátano Biosphere Reserve in Honduras.

5.2.1.3 End protected area size

Having large, protected, end point make the climate corridors more resilient to climate change. The graph below plots the area of end points by country. Panama’s Darien and Costa Rica’s La Amistad are the largest, high elevation, contiguous protected areas in central america. El Salvador, Guatemala and Belize lack large high elevation protected areas.

5.2.1.4 Multinational connectivity

The plot below shows the number of jurisdictional boundarys the climate corridors pass through. Most start and end within the same country (~1). The only exception is Belize, which has no high elevation protected areas of its own. Thus, the paths typically cross into a neighboring country.

5.2.1.5 Least cost path lengths

Short climate corridors will cost less to restore and maintain, longer corridors may be highly logistically complicated. The graph below shows the average LCP length, from start to finish for each corridor. Given the lack of high elevation areas within its borders, Belize has the longest corridors. Costa Rica has, on average, the shortest although Guatemala, Honduras and Panama are similar.

5.2.1.6 Average conductance

Paths we on average, less resistant to animal movement will be more effect as they should be easier to move through. The plot below shows the breakdown by country. Panama appears to be the least disturbed (most conductive) landscape, whereas the habitats in El Salvador, on average, traverse some highly unsuitable habitats.

5.2.1.7 Least-suitable habitat

It may be the case that the utility of a corridor is defined by the worst habitat within it. For example, the habitat is so poor that animals may not traverse it. We explored the minimum conductance of each climate corridor across the landscapes in the plot below. As expected, El Salvador performed very poorly owing to the landscape being highly human modified. Panamas worst scores were typiucally better than the average scores of most other countries.

5.2.1.8 Proportion of the path which is already protected

Understanding the proportion of these climate corridors which are fully protected can give insight into how resilient they are in the future. The plot below shows the proportion of each of the corridors designated as protected by country. Remarkably only 3 of the putative corridors are fully protected to date. These are corridors which terminate in Parque Nacional Cerro Hoya, Panama and Nombre de Dios, Honduras. Panama has several other corridors which are highly protected - including Darien and Reverendo Padre Jesús Héctor Gallego Herrera. Costa Rica’s highest protected climate corridor terminates in the Area de Conservación Guanacaste. As expected, El Salvador, on average, has the least prtected climate corridors. Over 20 putative corrdiors have <10% protection.

5.2.1.9 Proportion of forest cover

Regardless of whether the climate corridors are protected or not, they may have a reasonable amount of forest cover. The plot below shows the average forest cover per climate corridor by country. Panama has the highest average climate corridor forest cover (~90%), followed by Nicaragua. As expected, El Salvador had the lowest average forest cover (~60%).

5.2.2 Multivariate corridor prioritization

Given that there are multiple elements which must be balanced in order to prioritise locations for climate corridors (see above), one approach to integrating them is through the use of a principal component analysis (PCA). PCA’stake multiple input variables and decompose the correlated variation between them into “principal components” - which are essentially a mix of all the variables which load onto them. This allows users to simplify mutiple variables of interest into two or three interprtable axes.

Below, we perform this analysis on the characteristics provided above. Ccorridor length, start area, proportion forested, end area, average habitat conductance, and proportion of protected forest and their loading directions are shown as black lines, individual corridors are shown as points colored by their terminal country.

The graph shows that axis 1 is principally associated with increasing starting area size, forest quality and proportions protected as you move fromn left to right. Axis two is principally associated with path length, with climate corridors increasing in length from bottom to top.

5.2.2.1 The ‘sweet spot’

A sweet spot exists in the above plot which represents climate corridors which:

  • are not fully protected, but also have some protection
  • have large start and end locations
  • are which are not to long
  • are already situated in good habitat

These criteria all represent locations where climate corridor restoration initiatives would be suitable and, crucially, achievable. Using a green polygon below, we highlight the suite of corridor which fit this description:

Below we plot these 219 climate on a map to see if we can find a subset of wider landscapes which represent tartgets for climate corridor restoration.

We can also colour by score and look for patterns:

5.2.3 Conclusions

As you can see form the above map, the selected climate corridors are principally located in Panama, Costa Rica and Honduras. There is not much potential for restoration of climate corridors in El Salvador, and to a lesser extent Nicaragua, Guatamal and Belize.

The majority of corridors lie in and around La Amistad protected area, from both the Caribbean and Pacific directions, highlighting the importance of these regions.

Panama is relatively under represented as it already possess climate corridors with a cool percentage of protection and which contain good forest. These should be considered targets for preservation, not restoration per se.

5.3 Case study regions

Below we highlight the regions selected for the high resolution case studies in the following chapter.

And summarise the characteristics of the areas:

id land_area lowland_pa_area upland_pa_area protected_percent forest_percent mangrove_area elevation_min elevation_max n_corr avg_corr_length avg_corr_conductance
HondurasVolcan 19555.5 1595.4 64.6 14.2 27.9 686.0 0 2025 22 106.8 327.6
ElSalvador 28594.3 1520.0 848.5 10.4 39.3 454.3 0 2787 56 123.3 359.1
SelvaMaya 63755.8 22524.9 343.9 38.5 74.5 0.0 0 3799 67 179.4 574.6
Belize 50244.4 10550.8 216.5 30.7 78.1 801.5 0 2932 179 206.7 596.6
Mostikia 35319.9 17344.5 171.0 65.3 79.7 131.5 0 2261 21 193.4 678.8
IndioMaiz 18036.2 9332.1 541.8 58.6 74.8 7.2 0 3378 21 90.6 691.0
Guancaste 5638.9 737.3 33.1 24.8 56.3 10.6 0 1928 37 85.7 531.8
Amistosa 6310.8 1488.2 415.3 31.9 80.1 181.6 0 3451 13 100.1 721.5
Amistad 5211.7 874.0 1080.4 70.5 90.1 10.8 0 3393 7 37.7 685.8
Darien 13727.9 9492.9 42.3 90.3 94.1 174.8 0 1789 28 84.1 814.0